AI and the World's Hard Problems · Chemistry & materials · arXiv · 2026-10-19
A GENERIC-Informed Neural Network for Generative Rheology
GINNGER is a framework for discovering thermodynamically admissible constitutive equations from microstructural flow data. It learns the underlying thermodynamic generators rather than constitutive operators, embedding the GENERIC formalism so the first and second laws are satisfied by construction, and uses a spectral formulation in the eigenspace of the polymer conformation tensor so that shear, extensional and mixed flow data contribute to a single model. Validated on high-fidelity simulations, it accurately reconstructed constitutive behavior and predicted complex flow fields in previously unseen geometries.